The goal of auto-scaling is to apply Service Level Agreement (SLA) scaling policies to a Cloudbreak-managed HDP cluster.

The auto-scaling capability is based on Ambari Metrics and Ambari Alerts. Based on the blueprint used and the services running, Cloudbreak accesses all available metrics from the subsystem and defines alerts based on these metrics.

Enable Auto-scaling through Cloudbreak UI

Choose enable to enable auto-scaling:

Full size here.


Auto-scaling supports two alert types: metric and time based.

Metric-based Alerts

Metric-based alerts use Ambari metrics. These metrics have a default Threshold value configured in Ambari, which you can modify in Ambari web UI.

Change Default Threshold for an Ambari Metric

To change default threshold for an Ambari metric:

  1. Log in to Ambari web UI.
  2. From the header menu, select Alerts to open the Alerts page.
  3. Select an alert from the list.
  4. In the Configuration panel, click on Edit.
  5. Now you can modify the values in the Threshold section.

Full size here.

Create a New Metric-based Alert

To create a new Cloudbreak metric-based alert in the Cloudbreak web UI:

  1. Enter the alert name. Only alphanumeric characters (min 5, max 100 characters) are allowed.
  2. Enter a description for the new alert.
  3. Select a metric, and then its desired state. The Ambari metrics available are based on installed services and their state is based on the Ambari threshold value:
    • OK
    • WARN
  4. Enter the period (in minutes) to define the metric state endurance after the alert has been triggered. Only numeric characters are allowed.

Full size here.

Time-based Alerts

Time-based alerts are based on cron expressions, allowing alerts to be triggered based on time.

Create a Time-based Alert

To create a new Cloudbreak time-based alert in the Cloudbreak UI::

  1. Enter alert name. Only alphanumeric characters (min 5, max 100 characters) are allowed.
  2. Enter description for the new alert.
  3. Select a time zone for the new alert.
  4. Provide the cron expression to define the time-based job scheduler (cron expression) for this alert.

Full size here.

Scaling Policies

Scaling is the ability to increase or decrease the capacity of the HDP cluster or an application running on it based on an alert and according to the policy definition. After you set up your alerts and a scaling policy linked to them, Cloudbreak will execute the policy.

Scaling granularity is at the host group level; Thus you have an option to scale services or components only, not the whole cluster.

Create a New Scaling Policy

To create a new Cloudbreak scaling policy:

  1. Enter the policy name. Only alphanumeric characters (min 5, max 100 characters) are allowed.
  2. Select a type first, and then a value for the scaling adjustment:
    • node count - number of nodes (added or removed)
    • percentage - computed percentage adjustment based on the cluster size
    • exact - given size of the cluster
  3. Select the Ambari host group where the cluster is to be scaled.
  4. Select the previously created Cloudbreak alert to apply the scaling policy to it.

Full size here.

Cluster Scaling Configurations

An SLA scaling policy can contain multiple alerts. When an alert is triggered, a scaling adjustment is applied.

To make sure the scaling this adjustemnt doesn't oversize or undersize your cluster, you can keep the cluster size within defined boundaries using cluster size min. and cluster size max.

To avoid stressing the cluster, we have introduced a cooldown time period (in minutes). When an alert is raised and there is an associated scaling policy, the system will not apply the policy within the configured cooldown timeframe.

Note: In an SLA scaling policy the triggered rules are applied in order.

Full size here.

Explanation of the parameters:

Downscale Scaling Considerations

To keep your cluster healthy, Cloudbreak auto-scaling runs several background checks during downscale operation.

The API documentation was generated from the code using Swagger.

Edit on GitHub